In order to overcome the problem of low accuracy of the localization algorithm based on RSSI, an improved RSSI-based location algorithm is presented in this paper. To avoid the influence of the final positioning caused by error of a single RSSI measurement value, first, median model is used to correct each value of RSSI, then, corrected value of RSSI is used to estimate the relative size of the distance between network nodes which can communicate, finally eligible anchor node is used to calculate the position of the unknown node in this algorithm. The algorithm is simple, the nodes without increasing communication overhead during positioning and without hardware extension. Simulation results show that under circumstances of different path loss factor, positioning error of the node reduces the average 80% as compared to the original algorithm, greatly improving the accuracy of the node and greatly improves the accuracy of the node.